On Monday 25 January 2016 10:18:56 Kevin Hilman wrote:
Tyler Baker tyler.baker@linaro.org writes:
On 22 January 2016 at 05:19, Arnd Bergmann arnd@arndb.de wrote:
On Thursday 21 January 2016 22:14:26 kernelci. org bot wrote:
Warnings summary:
224 <stdin>:1307:2: warning: #warning syscall copy_file_range not implemented [-Wcpp]I don't think anyone has sent this so far, sending a patch now.
29 drivers/tty/serial/amba-pl011.c:190:27: warning: 'vendor_zte' defined but not used [-Wunused-variable]Patch sent a couple of days ago, got an Ack and waiting to be applied.
12 net/bluetooth/mgmt.c:5493:8: warning: 'r192' may be used uninitialized in this function [-Wmaybe-uninitialized] 12 net/bluetooth/mgmt.c:5493:8: warning: 'h192' may be used uninitialized in this function [-Wmaybe-uninitialized]This happens only in gcc-4.7 and gcc-4.8
5 crypto/wp512.c:987:1: warning: the frame size of 1168 bytes is larger than 1024 bytes [-Wframe-larger-than=] 1 crypto/wp512.c:987:1: warning: the frame size of 1112 bytes is larger than 1024 bytes [-Wframe-larger-than=]and this one only in gcc-4.7
Any chance we can update the toolchain on kernelci.org to 4.9 or higher? I'm normally testing with gcc-5.2, which has a little better warnings.
I'm ok with upgrading to 4.9 or higher. Kevin any issues with this?
No objection in principle, but we probably can't do this across the board without fallout.
I suspect we'll have issues with older stable kernels that have never been compiled with > 4.7, but I haven't actually tried.
My guess is that there is very little of that. We had one issues with gcc-4.8 through 4.8.2 generating bad code, but aside of that I don't remember a recent case where a newer compiler broke stuff.
There might be a couple of extra warnings in older kernels, but I've tried to ensure that all new warnings that got introduced from more recent gcc versions are fixed mainline kernels.
Arnd